Kafka indexing task duration and segment granularity

I’ve been playing around with this task type lately and I’m curious how Druid deals with segment sizes being longer than the task. For instance, if I want the task to run for 30 minutes but the segment granularity is set to HOUR, what happens to the segment during the handoff? It can’t be updated once published, so is the data just dropped when the next task starts up in the same segment?